Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: inconsistent do-file execution PLUS


From   David Jacobs <jacobs.184@sociology.osu.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: inconsistent do-file execution PLUS
Date   Mon, 09 Jun 2008 12:08:22 -0400

In support of Rufus' claims, I've experienced a less severe version of the same problem, but I think only with Stata version 7 (IC) for Windows.

After a long session, Stata would suddenly stop recognizing what I typed in the command window, and instead substitute characters I hadn't typed.

But to the best of my knowledge Stata didn't change my data. As with Rufus, if I quit Stata and then immediately relaunched it, the problem would go away at least until I again used the program for long time.

It is interesting that an apparently extreme version of this problem appears in Stata 9 for Macs.

Dave Jacobs

At 08:24 PM 6/8/2008, you wrote:

Friedrich (and others),

I think I found the problem.  I didn't have an ID variable for a
certain subsection of the data, but I'm still curious why the file
would execute differently each time.  The problem was the commands:

sort gameid
gen runavg_opp_pred=runavg_tm_pred[_n-1] if gameid[_n-1]==gameid
replace runavg_opp_pred=runavg_tm_pred[_n+1] if gameid[_n+1]==gameid
I'm using a dataset with many observations from years 2001-2007.  The
variable "gameid" was missing for all 2006 & 2007 observations.  Would
the sort execute differently each time for observations where the
sorting variable is missing?  That doesn't seem to make sense to me.


Another question that's somewhat related:

I just got Stata 10, but I had Stata 9 IC for Mac for the previous
year and every so often Stata would go bonkers on me.  I would execute
a simple command (like summarize) and it would give me different
results each time.  Sometimes Stata would even change my dataset by
deleting lots of observations and copying some rows hundreds of
times.  Other times, it would make the dataset look like nonsense (it
would be filled with characters such as ~øå¬πø¨©.  When I
quit Stata and restarted it, everything would be back to normal.  I
should note that Stata would usually only mess up like this after I
had been using it for awhile--it would never do this right after I
launched it.  Would have written about this problem earlier, but just
found out about the listserv.

Thanks again,

Rufus



On Jun 8, 2008, at 7:54 PM, Friedrich Huebler wrote:

Rufus,

At which step do you get different results? Please provide a more
precise description of your problem.

Thanks,

Friedrich

On Sun, Jun 8, 2008 at 7:48 PM, Rufus Peabody
<rufus.peabody@gmail.com> wrote:
Hey all,

I'm using Stata 10 SE for Mac.  I'm having a problem with the
execution of a
do-file--each time I do the do-file, I get different results for
only a
subset of the data.  First of all, is there any reason why running
the same
do-file which is using the exact same data would give me different
results?
The do-file is a bit long but I'll paste it in here, although I'm
not sure
how it would give me different results since I'm not simulating
anything...

Thanks,
Rufus


use iteration2a_lag_bet.dta


regress kpct lkpct_norm_pit_003 lkpct_norm_opp_01 ltotal_bf
lp_pa_norm_pit_up003 lstrike_pct_norm_opp_up01 lbip_pct_norm_opp_up01
kpct_int_bf if year<2005 & home==1
predict kpct_pred_home if home==1
regress kpct lkpct_norm_pit_003 lkpct_norm_opp_01 ltotal_bf
lp_pa_norm_pit_up003  lbip_pct_norm_opp_up01 kpct_int_bf if
year<2005 &
home==0
predict kpct_pred_away if home==0

regress bbpct ltotal_bf lbbpct_norm_pit_003 lbbpct_norm_opp_up01
bbpct_int_bf lp_pa_norm_pit_up003 lp_pa_norm_opp_up01
lstrike_pct_norm_opp_up01 lstrike_pct_norm_pit_up003
lhrpct_norm_opp_up01 if
year<2005 & home==1
predict bbpct_pred_home if home==1
regress bbpct ltotal_bf lbbpct_norm_pit_003 lbbpct_norm_opp_up01
bbpct_int_bf lp_pa_norm_pit_up003 lp_pa_norm_opp_up01
lstrike_pct_norm_opp_up01 lstrike_pct_norm_pit_up003 if year<2005 &
home==0
predict bbpct_pred_away if home==0

regress hrpct ltotal_bf hr_bip_pfbf ladjhrpct_norm_pit_up003
ladjhrpct_norm_opp_up01 lebh_bip_norm_opp_up01
lebh_bip_norm_pit_up003
lbip_pct_norm_pit_up003 lgdppct_norm_pit_up003 if year<2005 & home==1
predict hrpct_pred_home if home==1
regress hrpct ltotal_bf hr_bip_pfbf ladjhrpct_norm_pit_up003
ladjhrpct_norm_opp_up01 lebh_bip_norm_pit_003 lbip_pct_norm_pit_up003
lgdppct_norm_pit_up003 lgdppct_norm_opp_01 if year<2005 & home==0
predict hrpct_pred_away if home==0

regress babip babip_pfbf ltotal_bf lbabip_norm_pit_003
lbabip_norm_opp
lbip_pct_norm_pit_up003 if home==1 & year<2005
predict babip_pred_home if home==1
regress babip babip_pfbf ltotal_bf lbabip_norm_pit_003
lbabip_norm_opp_up
lbip_pct_norm_pit_up003 if home==0 & year<2005
predict babip_pred_away if home==0

regress woba woba_pfbf kpct_pred_home bbpct_pred_home hrpct_pred_home
babip_pred_home if home==1 & year<2005, robust
predict woba_pred_home if home==1
regress woba woba_pfbf kpct_pred_away bbpct_pred_away hrpct_pred_away
babip_pred_away if home==0 & year<2005, robust
predict woba_pred_away if home==0

regress ip lip_norm_pit_up003 ltotal_bf woba_pred_home
bbpct_pred_home if
year<2005 & home==1, robust
predict ip_pred_home if home==1
regress ip lip_norm_pit_up003 ltotal_bf woba_pred_away if year<2005 &
home==0, robust
predict ip_pred_away if home==0

regress bip_pct lbip_pct_norm_pit_up003 lkpct_norm_opp_up01
lbbpct_norm_opp_up01   if year<2005 & home==1, robust
predict bip_pct_pred_home if home==1
regress bip_pct lbip_pct_norm_pit_up003 lkpct_norm_opp_up01
lbbpct_norm_opp_up01   if year<2005 & home==0, robust
predict bip_pct_pred_away if home==0


regress kpct_pen lkpct_norm_pen_up01 lkpct_norm_opp_up01
lbip_pct_norm_opp_up01 ip_pred_home pitchers_lastweek_pen if
year<2005 &
home==1, robust
predict kpct_pen_pred_home if home==1
regress kpct_pen lkpct_norm_pen_01 lkpct_norm_opp_01
lbip_pct_norm_opp_up01
lbip_pct_norm_pen_01 lstrike_pct_norm_pen_01 pitchers_lastweek_pen
ip_pred_away if year<2005 & home==0, robust
predict kpct_pen_pred_away if home==0

regress bbpct_pen lbbpct_norm_pen_01  lbbpct_norm_opp_01
lstrike_pct_norm_pen_01 pitchers_lastweek_pen if year<2005 &
home==1, robust
predict bbpct_pen_pred_home if home==1
regress bbpct_pen  lbbpct_norm_opp_01 lstrike_pct_norm_pen_01
pitchers_lastweek_pen if year<2005 & home==0, robust
predict bbpct_pen_pred_away if home==0

regress babip_pen babip_pfbf lbabip_norm_pen_01  ip_pred_home
ip_lastweek_pen if year<2005 & home==1, robust
predict babip_pen_pred_home if home==1
regress babip_pen babip_pfbf lbabip_norm_pen_01  ip_pred_away
pitchers_lastweek_pen if year<2005 & home==0, robust
predict babip_pen_pred_away if home==0

regress hrpct_pen hrpct_pfbf ladjhrpct_norm_opp_01
ladjhrpct_norm_pen_01
ip_pred_home ip_lastweek_pen  if year<2005 & home==1
predict hrpct_pen_pred_home if home==1
regress hrpct_pen hrpct_pfbf ladjhrpct_norm_opp_01
ladjhrpct_norm_pen_01  if
year<2005 & home==0
predict hrpct_pen_pred_away if home==0

regress woba_pen kpct_pen_pred_home bbpct_pen_pred_home
babip_pen_pred_home
hrpct_pen_pred_home if year<2005 & home==1, robust
predict woba_pen_pred_home if home==1
regress woba_pen kpct_pen_pred_away bbpct_pen_pred_away
babip_pen_pred_away
hrpct_pen_pred_away if year<2005 & home==0, robust
predict woba_pen_pred_away if home==0


regress runavg_pen woba_pen_pred_home if year<2005 & home==1, robust
predict runavg_pen_pred_home if home==1
regress runavg_pen woba_pen_pred_away if year<2005 & home==0, robust
predict runavg_pen_pred_away if home==0


regress runavg woba_pred_home if year<2005 & home==1, robust
predict runavg_pred_home if home==1
regress runavg woba_pred_away if year<2005 & home==0, robust
predict runavg_pred_away if home==0

gen runavg_pred_home_IP=runavg_pred_home*ip_pred_home
gen runavg_pred_away_IP=runavg_pred_away*ip_pred_away

regress runavg_tm runavg_pen_pred_home runavg_pred_home
runavg_pred_home_IP
if year<2005 & home==1, robust beta
predict runavg_tm_pred_home if home==1
regress runavg_tm runavg_pen_pred_away runavg_pred_away
runavg_pred_away_IP
if year<2005 & home==0, robust beta
predict runavg_tm_pred_away if home==0

gen runavg_tm_pred=runavg_tm_pred_home if home==1
replace runavg_tm_pred=runavg_tm_pred_away if home==0

sort gameid
gen runavg_opp_pred=runavg_tm_pred[_n-1] if gameid[_n-1]==gameid
replace runavg_opp_pred=runavg_tm_pred[_n+1] if gameid[_n+1]==gameid

logit tm_win runavg_opp_pred runavg_tm_pred if home==1 & year<2005,
robust
predict home_win_new2 if home==1


gen away_win_new2=1-home_win_new2

gen strat1_ols_in_new2=tm_unit if home==1 & home_win_new2>tm_perc &
home_win_new2~=. & year<2005
gen strat1_ols_out_new2=tm_unit if home==1 & home_win_new2>tm_perc &
home_win_new2~=. & year>2004
replace strat1_ols_in_new2=opp_unit if home==1 &
away_win_new2>opp_perc &
home_win_new2~=. & year<2005
replace strat1_ols_out_new2=opp_unit if home==1 &
away_win_new2>opp_perc &
home_win_new2~=. & year>2004


regress tm_unit home_win_new2 tm_perc if home==1 & home_win_new2~=. &
year<2005
predict exp_return_home_new2 if home_win_new2!=.
regress opp_unit home_win_new2 tm_perc if home==1 &
home_win_new2~=. &
year<2005
predict exp_return_away_new2 if home_win_new2!=.

***
gen strat2_ols_in_new2=opp_unit if exp_return_away_new2>0 &
away_win_new2>opp_perc & year<=2004 & home_win_new2~=. & home==1 &
exp_return_away_new2!=.
replace strat2_ols_in_new2=tm_unit if exp_return_home_new2>0 &
home_win_new2>tm_perc & year<=2004 & home_win_new2~=. & home==1 &
exp_return_home_new2!=.
gen strat2_ols_out_new2=opp_unit if exp_return_away_new2>0 &
away_win_new2>opp_perc & year>2004 & home_win_new2~=. & home==1 &
exp_return_away_new2!=.
replace strat2_ols_out_new2=tm_unit if exp_return_home_new2>0 &
home_win_new2>tm_perc & year>2004 & home_win_new2~=. & home==1 &
exp_return_home_new2!=.


bysort year: summ strat2_ols_in strat2_ols_out strat1_ols_in
strat1_ols_out
[iweight=abs(home_win_new2-tm_perc)]*
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/


*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index